Introducing a computationally light approach to estimate forest height and fractional canopy cover from Sentinel-2 data

Abstract

Accurate and timely monitoring of forest attributes such as canopy height and cover is crucial in the face of global climate change and its impact on forests, in particular across semi-arid and fragile ecosystems. Motivated by this, we explored the potential of Sentinel-2 satellite imagery for estimating forest height (H) and fractional canopy cover (FCC) in semi-arid mountainous ecosystems, as challenging cases for optical remote sensing due to the prominent influence of soil background. Our investigation tackles a gap in current research by utilizing multi-temporal data to understand the effects of tree growth on Sentinel-2 spectral bands, and unravelling the spectral content of Sentinel-2 to estimate H and FCC at 10 m spatial resolution. We first estimated FCC and then H using a newly proposed vegetation index (VI) by optimizing Sentinel-2 band coefficients to mitigate soil influence. FCC was used included in the model to estimate H. We estimated the change in H (ΔH) over one and two-year growth periods, analysing the relationship between spectral variations and tree height growth. We propose a workflow for optimizing our index for FCC, H and ΔH to address the challenge of soil background influence in semi-arid tree cover. The results indicate a strong relationship between the optimized index and FCC, H and ΔH, with R² = 0.76, 0.70 and 0.62 respectively obtained using a linear model. The proposed method additionally underwent rigorous testing in a temperate forest study area characterized by dense forest stands and reference data from laser scanning, and returned moderate results which indicate the robustness and reliability of our approach to estimate H across forest structures.